Peltier Chad, Becker Mark W
Department of Psychology, Michigan State University.
J Exp Psychol Hum Percept Perform. 2016 Sep;42(9):1466-76. doi: 10.1037/xhp0000248. Epub 2016 May 5.
The probability of missing a target increases in low target prevalence search tasks. Wolfe and Van Wert (2010) propose 2 causes of this effect: reducing the quitting threshold, and conservatively shifting the decision making criterion used to evaluate each item. Reducing the quitting threshold predicts that target absent responses will be made without fully inspecting the display, increasing misses due to never inspecting the target (selection errors). The shift in decision criterion increases the likelihood of failing to recognize an inspected target (identification errors). Though there is robust evidence that target prevalence rates shift quitting thresholds, the proposed shift in decision making criterion has little support. In Experiment 1 we eye-tracked participants during searches of high, medium, and low prevalence. Eye movements were used to classify misses as selection or identification errors. Identification errors increased as prevalence decreased, supporting the claim that decision criterion becomes more conservative as prevalence decreases. In addition, as prevalence decreased, the dwell time on targets increased while dwell times on distractors decreased. We propose that the effect of prevalence on decision making for individual items is best modeled as a shift in criterion in a drift diffusion model, rather than signal detection, as drift diffusion accounts for this pattern of decision times. In Experiment 2 we replicate these findings while presenting stimuli in an rapid serial visual presentation (RSVP) stream. Experiments 1 and 2 were consistent with the conclusion that prevalence rate influences the item-by-item decision criterion, and are consistent with a drift diffusion model of this decision process. (PsycINFO Database Record
在低目标出现率的搜索任务中，错过目标的概率会增加。沃尔夫和范·韦特（2010年）提出了造成这种效应的两个原因：降低停止阈值，以及保守地改变用于评估每个项目的决策标准。降低停止阈值预示着在没有充分检查显示的情况下就会做出目标不存在的反应，这会因从未检查目标（选择错误）而增加漏报情况。决策标准的改变增加了未能识别已检查目标的可能性（识别错误）。尽管有确凿证据表明目标出现率会改变停止阈值，但所提出的决策标准的改变却几乎没有得到支持。在实验1中，我们在高、中、低出现率的搜索过程中对参与者进行了眼动追踪。眼动被用来将漏报分类为选择错误或识别错误。随着出现率的降低，识别错误增加，这支持了随着出现率降低决策标准会变得更加保守的观点。此外，随着出现率降低，对目标的注视时间增加，而对干扰项的注视时间减少。我们提出，出现率对单个项目决策的影响最好被建模为漂移扩散模型中标准的改变，而不是信号检测，因为漂移扩散模型能够解释这种决策时间模式。在实验2中，我们在快速序列视觉呈现（RSVP）流中呈现刺激时重复了这些发现。实验1和实验2都与出现率影响逐个项目的决策标准这一结论一致，并且与该决策过程的漂移扩散模型一致。（《心理学文摘数据库记录》 ）
